Scholarlink

This study advances the field of structural engineering by developing a machine-learning model for evaluating the factors affecting the lateral stiffness of unbounded post-tensioned cross-laminated timber (CLT) shear walls. In order to do that, the author has established a precise stiffness model for post-tension cross-laminated timber shear walls, which is explored by developing and validating a finite element database. A comprehensive dataset covering various post-tension cross-laminated timber parameters is meticulously created, comprising 137 distinct model samples undergoing rigorous simulations and analyses. Machine learning models, including Random Forest and Categorical Boosting (CatBoost), are extensively evaluated, with the Random Forest model emerging as the superior performer, exhibiting higher accuracy and lower error metrics. Further insights into model predictions are gleaned through SHapley Additive explanations (SHAP) analysis, revealing the relative importance of input features and identifying that shear wall width and thickness are the most critical factors affecting the lateral stiffness.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning Techniques for Analysing Post-tensioned Cross-Laminated Timber Behaviour Under Lateral Loads

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • K. G. M. Kandethanthri,
  • Ghazanfarah Hafeez

摘要

This study advances the field of structural engineering by developing a machine-learning model for evaluating the factors affecting the lateral stiffness of unbounded post-tensioned cross-laminated timber (CLT) shear walls. In order to do that, the author has established a precise stiffness model for post-tension cross-laminated timber shear walls, which is explored by developing and validating a finite element database. A comprehensive dataset covering various post-tension cross-laminated timber parameters is meticulously created, comprising 137 distinct model samples undergoing rigorous simulations and analyses. Machine learning models, including Random Forest and Categorical Boosting (CatBoost), are extensively evaluated, with the Random Forest model emerging as the superior performer, exhibiting higher accuracy and lower error metrics. Further insights into model predictions are gleaned through SHapley Additive explanations (SHAP) analysis, revealing the relative importance of input features and identifying that shear wall width and thickness are the most critical factors affecting the lateral stiffness.